Viral Content Propagation Analyzer in a Social Networking System

1. A computer-implemented method for improving computer-implemented content selection, the method comprising: processing user activities in a social networking system through a relevancy filter to identify a subset of user activities that are relevant to a viral propagation study; in response to selecting a user activity as the graph exploration seed, constructing a user activity cascade by exploring a social graph in the social networking system, starting from a starter social network node corresponding to the selected user activity, wherein the user activity cascade comprises social network nodes found during said exploring; and determining that the user activity cascade is virally propagating, wherein the determining improves computer-implemented content selection and wherein the determining is based on one or more of: a total size of the user activity cascade, an expansion rate of the user activity cascade, a diversity measure of the user activity cascade, or any combination thereof; wherein, one or more of: said exploring the social graph includes exploring one or more social network nodes connected to one or more already explored social network nodes via one or more social network edges and wherein the social network edges correspond to one or more specific social connection types; said determining that the user activity cascade is virally propagating includes counting a number of social network nodes in the user activity cascade corresponding to user activities within a time window; said exploring includes exploring one or more social network nodes connected via directional social network edges, and wherein identifying the root network node includes exploring the social graph along one of the directions until no further social network node is found; the method further comprises: determining whether the user activity cascade is similar to another user activity cascade being monitored by a viral analysis system; and in response to determining that the user activity cascade is similar to the other user activity cascade, merging the user activity; or any combination thereof.

2. The computer-implemented method of claim 1 , wherein said processing the user activities includes confirming a newly submitted user activity in the social networking system is relevant via the relevancy filter; and wherein the method includes sampling the subset of user activities from the relevancy filter by, in response to confirming the newly submitted user activity is relevant, determining whether to select the newly submitted user activity as the graph exploration seed according to a random or periodic determination.

3. The computer-implemented method of claim 1 , wherein the one or more specific social connection types include a content sharing edge type that describes a first social networking node being a re-share of a second social networking node.

4. The computer-implemented method of claim 1 , wherein said exploring stops when no more social network node is found or when a social network node is detected as being part of another user activity cascade being monitored by a viral analysis system.

5. The computer-implemented method of claim 1 , wherein said exploring includes exploring only one or more social network nodes corresponding to content publication activities.

6. The computer-implemented method of claim 1 , further comprising identifying a root social network node in the user activity cascade as corresponding to a source content object that started to virally propagate.

7. The computer-implemented method of claim 1 , wherein said determining whether the user activity cascade is similar to the other user activity cascade includes comparing a first set of content objects involved in the user activity cascade against a second set of content objects involved in the other user activity Cascade in terms of overlap in title, links, comments, content text, images, or any combination thereof.

8. The computer-implemented method of claim 7 , wherein said comparing includes translating at least a portion of the first set of content objects or a portion of the second set of content objects to utilize a single language for comparison.

9. A computer readable data memory storing computer-executable instructions that, when executed by a computer system, cause the computer system to perform a computer-implemented method for improving computer-implemented content management using a dashboard, the method comprising: processing a user activity in a social networking system through a relevancy filter to determine whether the user activity is relevant to a viral propagation study; when the user activity is determined to be relevant to the viral propagation study, constructing a user activity cascade by exploring a social graph in the social networking system, starting from a social object node corresponding to the user activity, wherein the user activity cascade comprises social network nodes found during said exploring; determining that the user activity cascade is virally propagating based at least upon a total size of the user activity cascade; and generating the dashboard to monitor the user activity cascades, thereby improving computer-implemented content management; wherein, one or more of: the method further comprises: computing a virality measuring statistic of the user activity cascade; and displaying the virality measuring statistical on the dashboard; the method further comprises: compiling a demographic profile of the user activity cascade; and displaying the demographic profile on the dashboard; the method further comprises: analyzing user activities in the user activity cascade to generate a summary expression; and displaying the summary expression on the dashboard; or any combination thereof.

10. The computer readable data memory of claim 9 , wherein the method further comprises: computing a predictive trend of user activities in the user activity cascade; and displaying the predictive trend on the dashboard.

11. The computer readable data memory of claim 9 , wherein the method further comprises: performing content analysis on content objects associated with social network nodes in the user activity cascade to determine a content attribute; and displaying the content attribute on the dashboard.

12. A social networking system for improving computer-implemented content management, the system comprising: a sampler engine configured to receive user activities from a relevancy filter and randomly or periodically select the user activity for the graph exploration engine as a graph exploration seed; a graph exploration engine configured to construct a user activity cascade by exploring a social graph in the social networking system, starting from a social object node corresponding to a user activity indicated as relevant for a viral propagation study, wherein the user activity cascade comprises social network nodes found during said exploring; a cascade repository configured to maintain an index of one or more user activity cascades into the user activity cascade; and a virality determination engine configured to determine that the user activity cascade is virally propagating based on one or more of: a total size of the user activity cascade, an expansion rate of the user activity cascade, a diversity measure of the user activity cascade, or any combination thereof, wherein the determination that the user activity cascade is virally propagating provides the improvement in computer-implemented content management.

13. The social networking system of claim 12 , further comprising a cascade analysis engine configured to compute a predictive trend associated with the user activity cascade and to provide the predictive trend to an application service of the social networking system.